After a lengthy pause caused by regulatory hurdles, Meta will begin training its AI models using public content shared by adult UK citizens on Facebook and Instagram.
Tech giant Meta, the company behind popular social media – Facebook and Instagram, resumes generative AI training in the UK, with regulatory feedback incorporated into the process for more transparency.
Meta has already released a few iterations of its large language model (LLM) called Llama. It can converse with users in eight languages, write computer code and solve complex mathematical problems. Besides, the company launched a range of various generative AI models that can generate text and images and detect AI-generated speech within larger audio snippets.
At the same time, the rollout of these products in Europe was postponed in June due to regulatory concerns. A few privacy regulators questioned Meta’s plan to harness user data. While the company claims that only publicly available and licensed online data will be used, the Irish watchdog expressed concern that confidential information of account holders on social media platforms such as Instagram and Facebook could be involved in the AI training process.
Meta has accused the EU of holding back AI progress by refusing to allow EU citizens’ posts to be used for AI training. The plan is still on pause for the EU countries. Nevertheless, publicly shared posts of the UK users of Facebook and Instagram will now be used as part of the dataset for Meta’s AI models.
Publicly available information includes public posts, comments, photos and captions. It doesn’t include private messages or any information from accounts of UK citizens under the age of 18.
In the company’s opinion, it will help bring AI-driven features within Meta products to the UK much sooner. Besides, Meta claims that this way its generative AI models will reflect British culture, history and idiom.
Resuming the AI training plan is possible due to Meta’s fruitful cooperation with the UK’s Information Commissioner’s Office (ICO). The company praises the constructive approach this regulator has chosen.
A mutual consensus was reached in terms of enhanced transparency. In particular, now UK users of Facebook and Instagram will receive in-app notifications to explain how their data will be used for AI training and how they can access an objection form at any time. Due to feedback from the ICO, the objection form became simpler, more prominent and easier to find. In addition, users who objected to their data use once will not be asked about it again.
